AI critic says doom talk is a distraction
Archived — this story has rotated out of today’s deck. It is kept here in full.
The gist
Timnit Gebru says the "machine-god narrative" is "meant to distract us" from AI harms already happening.
The split matters because it shapes which AI risks get regulated and funded.
Background
Gebru, hired at Google in 2018 to evaluate bias in its AI tools, later left the company in a highly publicized departure and became one of the industry's most prominent critics. In a WIRED interview published September 11, 2026, she argued the "machine-god narrative" is "meant to distract us" from harms already occurring. The renewed debate was set off by the public resignation of Anthropic researcher Jacob Coxon, who accused his employer and OpenAI of "gambling with our lives" in the race to build advanced models.
How it unfolded
- 2018Timnit Gebru was hired at Google to evaluate biases within the company's fast-advancing AI tools.
- date unknownGebru and fellow researchers presented a paper on the potential dangers of probabilistic large language models; Google rejected it.
- Tuesday (date unknown)Anthropic researcher Jacob Coxon publicly resigned, accusing his employer and OpenAI of "gambling with our lives" in a race to build advanced models.
- Sep 11, 2026WIRED published an interview in which Gebru said the "machine-god narrative" is "meant to distract us"; BigGo Finance reported the AI doom debate split, with critics calling extinction talk a distraction from real harms.
Who’s saying what
- Critic
- Gebru says the "machine-god narrative" is "meant to distract us" from harms already happening, and that what's really existential is AI powering autonomous weapons and killing machines actually used in warfare, plus climate risks and bosses using AI as an excuse to cut workers.
- Researcher
- Jacob Coxon, who resigned from Anthropic, says advanced systems could become "smart enough to kill us" and accused Anthropic and OpenAI of "gambling with our lives" in a race to build advanced models.
- Analysts
- Business Insider asked four leading chatbots to rank AI doomsday scenarios, and all dismissed the "killer" scenario.
Still unverified
A colleague of Coxon, responding online, said people at Anthropic "really do earnestly believe AI could kill all humans" and gave a personal estimate of ">10% within the next decade"; both the claim and the estimate are attributed to an unnamed colleague posting online.